Free Radical Biology and Medicine 59 (2013) 69–84

Contents lists available at SciVerse ScienceDirect

Free Radical Biology and Medicine

0891-58

http://d

Abbre

acid; CT

tion for

tion; FT

farneso

mance l

chroma

reaction

fragmen

time-of

selectedn Corr

E-m

journal homepage: www.elsevier.com/locate/freeradbiomed

Methods in Free Radical Biology and Medicine

Analytical strategies for characterization of oxysterol lipidomes:Liver X receptor ligands in plasma

William J. Griffiths a,n, Peter J. Crick a, Yuchen Wang b, Michael Ogundare a, Karin Tuschl c,Andrew A. Morris d, Brian W. Bigger d, Peter T. Clayton c, Yuqin Wang a

a Institute of Mass Spectrometry, College of Medicine, Swansea University, Swansea SA2 8PP, UKb Clinical Laboratory, Jinan Infectious Disease Hospital, Shandong University, Jinan, Shandong, Chinac Clinical and Molecular Genetics Unit, Institute of Child Health, University College London, London WC1N 1EH, UKd Willink Biochemical Genetics Unit, Genetic Medicine, St Mary’s Hospital, CMFT, Manchester M13 6WL, UK

a r t i c l e i n f o

Available online 27 July 2012

Keywords:

Lipidomics

Sterol

Steroid

Oxysterol

Liquid chromatography

Mass spectrometry

Bile acid

Derivatization

Cerebrotendinous xanthomatosis

Nuclear receptor

G-protein-coupled receptor

Liver X receptor

Free radicals

49 & 2012 Elsevier Inc.

x.doi.org/10.1016/j.freeradbiomed.2012.07.02

viations: API, atmospheric pressure ionizatio

X, cerebrotendinous xanthomatosis; EADSA,

sterol analysis; DHEA, dehydroepiandrostero

, Fourier transform; FWHM, full width at hal

id X receptor; GC, gas chromatography; GP, G

iquid chromatography; HSD, hydroxysteroid

tography; LIT, linear ion trap; LXR, liver X rec

monitoring; MS, mass spectrometry or spec

tation; PQD, pulsed Q collision induced disso

-flight; RIC, reconstructed ion chromatogram

ion recording; SPE, solid-phase extraction

esponding author. Fax: þ44 1792 295554.

ail address: w.j.griffiths@swansea.ac.uk (W.J.

Open access under C

a b s t r a c t

Bile acids, bile alcohols, and hormonal steroids represent the ultimate biologically active products of

cholesterol metabolism in vertebrates. However, intermediates in their formation, including oxysterols

and cholestenoic acids, also possess known, e.g., as ligands to nuclear and G-protein-coupled receptors,

and unknown regulatory activities. The potential diversity of molecules originating from the cholesterol

structure is very broad and their abundance in biological materials ranges over several orders of

magnitude. Here we describe the application of enzyme-assisted derivatization for sterol analysis

(EADSA) in combination with liquid chromatography–electrospray ionization–mass spectrometry to

define the oxysterol and cholestenoic acid metabolomes of human plasma. Quantitative profiling of

adult plasma using EADSA leads to the detection of over 30 metabolites derived from cholesterol,

some of which are ligands to the nuclear receptors LXR, FXR, and pregnane X receptor or the

G-protein-coupled receptor Epstein–Barr virus-induced gene 2. The potential of the EADSA technique

in screening for inborn errors of cholesterol metabolism and biosynthesis is demonstrated by the

unique plasma profile of patients suffering from cerebrotendinous xanthomatosis. The analytical

methods described are easily adapted to the analysis of other biological fluids, including cerebrospinal

fluid, and also tissues, e.g., brain, in which nuclear and G-protein-coupled receptors may have important

regulatory roles.

& 2012 Elsevier Inc. Open access under CC BY license.

Introduction

Oxysterols and their downstream metabolites, including choles-tenoic acids, represent important biologically active components ofplasma. These molecules are of increasing interest to bioscientists onaccount of their important signaling roles in the immune system[1–3], as agonists to nuclear receptors [4–7], and as markers of

7

n; CDCA, chenodeoxycholic

enzyme-assisted derivatiza-

ne; ESI, electrospray ioniza-

f-maximum height; FXR,

irard P; HPLC, high-perfor-

dehydrogenase; LC, liquid

eptor; MRM, multiple

trometer; MSn, multistage

ciation; Q-TOF, quadrupole–

; RP, reversed phase; SIR,

Griffiths).

C BY license.

oxidative stress [8], atherosclerosis [9,10], and neurodegenerativedisease [11]. Oxysterols can be formed from cholesterol andits sterol precursors both enzymatically and nonenzymatically.In vertebrates, the first step of all cholesterol metabolism leads tothe formation of an oxysterol. 22R-Hydroxycholesterol and 20R,22R-dihydroxycholesterol are the precursors of pregnenolone and steroidhormones [12], whereas 7a-, 24S-, 25-, and (25R),26-hydroxycholes-terols all represent precursors of bile acids [13]. Some of theseoxysterols and their downstream metabolites are ligands to nuclearreceptors, e.g., liver X receptors (LXRs)1 [5,6], farnesoid X receptor(FXR) [7], pregnane X receptor [4,14], vitamin D receptor [15], andother receptors involved in lipid homeostasis, e.g., INSIG [16,17], andalso G-protein-coupled receptors [2,18,19]. Oxysterols also play a rolein the immune response [1,20,21], in which, e.g., 25-hydroxycholes-terol is secreted by macrophages in response to Toll-like receptoractivation and suppresses immunoglobulin A production, whereas itsmetabolite 7a,25-dihydroxycholesterol directs B-cell migration [1,3].Clearly, the presence of these regulatory molecules in biological fluidsis of major physiological importance and analytical methods arerequired to reliably identify and quantify such molecules.

W.J. Griffiths et al. / Free Radical Biology and Medicine 59 (2013) 69–8470

Oxysterols have traditionally been analyzed in plasma by gaschromatography (GC)–mass spectrometry (MS), after saponificationand derivatization, utilizing selected ion recording (SIR) to attainthe necessary sensitivity [22], although liquid chromatography(LC)–MS methods utilizing SIR or multiple reaction monitoring(MRM) are gaining popularity [23–25]. To improve MS responsein LC–electrospray ionization (ESI)–MS studies a number of groupsare now exploiting derivatization methods to enhance ionization[26–29]. Cholestenoic acids are usually analyzed by GC–MS in ananalysis separate from oxysterols [30], this is on account of differentrequirements of sample preparation. However, Axelson and collea-gues developed a GC–MS assay for C27 acids in blood and plasma [31],which was extended to include some oxysterols.

Oxysterols represent just one subgroup of the lipidome, andin recent years lipidomics has become a field of great activity in,and intense interest to, the bioscience community [32–35].Many of the lipidomic studies performed have been on biofluids,particularly plasma or serum [25]. The dominant technology hasbeen ESI-MS. ESI-MS analysis has been performed linked to LC,i.e., LC-ESI-MS, and also in a stand-alone direct infusion mode.Stand-alone, or shotgun, lipidomics offers the advantage ofsimplicity, but is unable to differentiate between isobaricmetabolites. Interfacing ESI with LC separation can overcomethis shortcoming, but introduces some added complexity tothe analysis. GC–MS offers an alternative analytical method;however, the requirements of solvolysis and/or hydrolysisfollowed by derivatization, often of multiple functional groupswith freshly prepared reagents and solvents, discourage manywould-be analysts [36].

Any lipidomic experiment consists of essentially three keystages: (i) lipid extraction, (ii) lipid analysis, and (iii) lipidquantification. Clearly the extremely different physical propertiesof many lipids and their presence in cells, tissues, and body fluidsat widely different levels make analysis of the global lipidomeextremely challenging. An alternative to global lipidomics is toadopt a targeted approach. Here a particular class of lipid istargeted, usually based on physicochemical properties. This is theapproach adopted by many lipid scientists [32,33,37]. Whereasglycerolipids, glycerophospholipids, and sphingolipids are oftenwell represented in global ESI-MS-based lipidomic studies, this isnot true of members of the other classes [38]. The explanation forthis is simple; most ESI-MS methods are biased toward the mostabundant and readily ionized compounds. Cholesterol and someof its metabolites are present at high levels in blood and plasmabut are barely detectable in a global lipidomics experiment basedon the ESI-MS analysis of these fluids. This is on account ofdifficulties experienced in sample handling and poor ionizationcharacteristics of sterols. Whereas established MS methods existfor analysis of the ultimate products of cholesterol metabolism,i.e., hormonal steroids and bile acids [39], methods for theanalysis of intermediates in their biosynthesis, i.e., oxysterolsand cholestenoic acids, are less mature, and to date this region ofthe lipidome has been largely ignored.

Over the past decade we have developed targeted methods forsterol, oxysterol, and cholestenoic acid analysis, which we haveexploited in the analysis of brain tissue, cerebrospinal fluid, plasma/serum, cells, and cell media [40–45]. Our methodology is based onseparating oxysterols and cholestenoic acids from cholesterol in thefirst step of the sample preparation process. This avoids thepotential problem of cholesterol autoxidation generating oxysterolsnonenzymatically with structures similar to those formed endogen-ously [46] and allows the subsequent storage of oxysterols withoutthe possibility of their formation via cholesterol and air. Oxysterolsare then activated to allow subsequent ‘‘click chemistry’’ with acharge-baring group, which enhances their ESI-MS response. In ourexperience, unprocessed samples can be stored effectively at �80 1C

before sample preparation with minimal autoxidation. However,growing peaks corresponding to 7-oxocholesterol, 7b-hydroxycho-lesterol, 5,6-epoxycholesterol, and cholestane-3b,5a,6b-triol shouldalert the analyst of potential autoxidation problems.

Note on nomenclature

Here we regard a steroid as a molecule based on the cyclo-pentanoperhydrophenanthrene ring structure. In vertebrateshormonal steroids usually contain 18, 19, or 21 carbon atoms;bile acids 24 carbon atoms with a carboxylic acid group at C-24;and cholestanoic and cholestenoic acids 27 carbon atoms with anacid group at C-26 or C-27. In cholestenoic acids the stereo centerat C-25 is usually 25R. Endogenous sterols are precursors ormetabolites of cholesterol and include cholestenoic acids, which,like cholesterol, possess a hydroxy (or oxo) group at C-3 andusually contain 27 carbons. Oxysterols are a category of sterols,mostly derived from cholesterol, containing an additional oxygenfunction. In this article we have adopted the nomenclaturerecommended by the Lipid Maps Consortium [37].

Principles

In this article we describe a LC-ESI-MS method for the quanti-tative profiling of a wide range of oxysterols and their downstreamacidic metabolites from microliter quantities of plasma. Our methodis based on ethanol extraction, separation of cholesterol metabolitesfrom cholesterol itself by reversed-phase (RP) solid-phase extraction(SPE), followed by enzyme-assisted derivatization for sterol analysis(EADSA). EADSA consists of enzymatic conversion of 3b-hydroxy-5-ene- and 3b-hydroxy-5a-hydrogen-containing sterols to 3-oxo-4-ene and 3-oxo sterols followed by tagging a positively chargedquaternary nitrogen group to the resulting oxo group in a ‘‘clickreaction’’ (Scheme 1). Analysis and quantification are performed byLC-ESI-MS. Our preference is to perform mass analysis at high massresolution (30,000, full width at half-maximum height, FWHM)using stable-isotope or structural analogue internal standards forquantification, with compound identification achieved using exactmass measurements (o5 ppm) and multistage fragmentation(MSn). We perform these analyses on an LTQ-Orbitrap instrument.Alternative MS formats can be used, such as quadrupole–time-of-flight (Q-TOF), tandem quadrupole, and cylindrical or linear ion trap(LIT), but none of these offers the combination of high-resolutionexact mass measurements and MSn provided by LTQ-Orbitrap orLTQ-FT-ICR instruments. We complement the EADSA LC-ESI-MSmethod with shotgun ESI-MS, also performed at high resolutionwith exact mass measurement and MS2, to characterize oxysterolsand their downstream metabolites conjugated with sulfuric and/orglucuronic acids. Shotgun ESI-MS is appropriate for analysis ofcholesterol metabolites sulfated at C-3 and thus inaccessible toEADSA. Using shotgun ESI-MS we do not attempt to exactly identifyor quantify conjugated oxysterols on account of an absence ofauthentic standards. Ongoing work is in progress to rectify thissituation by the synthesis of appropriate standards.

Materials

High-performance liquid chromatography (HPLC)-grade water,absolute ethanol, and other HPLC-grade solvents were from FisherScientific (Loughborough, UK) or Sigma–Aldrich (Dorset, UK). Aceticacid was AnalaR NORMAPUR grade (BDH, VWR, Lutterworth, UK).Authentic sterols, steroids, bile acids, and their precursors were fromAvanti Polar Lipids (Alabama, USA), Steraloids, Inc. (Rhode Island,USA), Sigma–Aldrich, or previous studies in our laboratories [41].

19 1211 1314 15

1617

18 20

21 22

23 24 2526

27

O

OH

O

OH

HO

Cholesterol Oxidase

GP Hydrazine

123 4 5 6 7

891014 15

OC27H44O3Exact Mass: 416.32904

C27H42O3Exact Mass: 414.31340

OO O

OH

O

OH

N

HNN

O

MS2N

HN

O

C29H45N2O3+

Exact Mass: 469.3425 -79.0422

C34H50N3O3+

Exact Mass: 548.38467

Scheme 1. Enzyme-assisted derivatization for sterol analysis. Sterols possessing a 3b-hydroxy-5-ene function are oxidized with cholesterol oxidase (from Streptomyces sp.),

and the resulting 3-oxo-4-ene group is then derivatized with Girard P (GP) hydrazine in a ‘‘click reaction’’ [42]. The cholesterol oxidase enzyme is also active toward sterols

with a 3b-hydroxy-5a-hydrogen function generating 3-oxo sterols [47]. Sterols naturally containing an oxo group can be derivatized with GP hydrazine in the absence of

cholesterol oxidase. Once derivatized, sterols are analyzed by LC-ESI-MS and MSn. The derivatization procedure is exemplified by 3b-hydroxycholest-(25R)-5-en-26-oic acid.

The nomenclature recommended by Lipid Maps has been adopted, by which a hydroxyl group attached to the terminal carbon of the cholestene side chain introducing

the R configuration at C-25 is said to be attached to C-26. This nomenclature is also extended to C27 acids [37].

Table 1Suppliers of reagents and materials.

Material Supplier Cat. No

24(R/S)-[26,26,26,27,27,27-2H6]Hydroxycholesterola Avanti Polar Lipids 700049P

24(R/S)-[25,26,26,26,27,27,27-2H7]Hydroxycholesterola Avanti Polar Lipids 700018P

http://avantilipids.com/index.php?option=com_content&view=article&id=2323&Itemid=305&catnumber=700018

[25,26,26,26,27,27,27-2H7]Cholesterol Avanti Polar Lipids 700041P

http://avantilipids.com/index.php?option=com_content&view=article&id=693&Itemid=305&catnumber=700041

Girard P reagent TCI Europe G0030

http://www.tcieurope.eu/en/catalog/G0030.html

Cholesterol oxidase from Streptomyces sp. Sigma–Aldrich C8868

http://www.sigmaaldrich.com/catalog/product/sigma/c8649?lang=en&region=GB

Certified Sep-Pak C18 200-mg, 3-ml, VAC cartridges Waters 186004618

http://www.waters.com/waters/partDetail.htm?partNumber=186004618

BD Biosciences Luer-lock syringe Sigma–Aldrich Z192120-100EA

http://www.sigmaaldrich.com/catalog/product/aldrich/z192120?lang=en&region=GB

Corning 15-ml centrifuge tube Fisher Scientific CFT-420-031Y

https://extranet.fisher.co.uk/insight2_uk/mainSearch.do

Corning 50-ml centrifuge tube Fisher Scientific CFT-900-031F

https://extranet.fisher.co.uk/insight2_uk/mainSearch.do

Greiner 12-ml round-bottom tubes, polypropylene Sigma–Aldrich Z642975

http://www.sigmaaldrich.com/catalog/product/sigma/z642975?lang=en&region=GB

Cultubes 4-ml polypropylene tube Simport Plastics T415-2A

http://www.bioventures.com/_pdf/Simport-Scientific-Catalog.pdf

Microcentrifuge tubes Fisher Scientific TUL-150-290U

https://extranet.fisher.co.uk/insight2_uk/getProduct.do?productCode=TUL-150-290U&resultSetPosition=26

a 24(R/S)-[26,26,26,27,27,27-2H6]Hydroxycholesterol is no longer commercially available and has been replaced by 24(R/S)-[25,26,26,26,27,27,27-2H7]hydroxycho-

lesterol.

W.J. Griffiths et al. / Free Radical Biology and Medicine 59 (2013) 69–84 71

Girard P (GP) reagent (1-(carboxymethyl)pyridinium chloride hydra-zide) was from TCI Europe (Oxford, UK) and cholesterol oxidasefrom Streptomyces sp. was from Sigma–Aldrich. Certified Sep-Pak C18

200-mg cartridges were from Waters (Elstree, UK). Luer-lock

syringes were from BD Biosciences (Sigma–Aldrich). All plasticmaterials used were made of polypropylene. Latex-containingmaterial, including gloves, should be avoided. Further supplierdetails including Web addresses are given in Table 1.

Plasma

C18SPE-1

SPE-1-Fr-1,70% EtOH

Cholesterol oxidase (inphosphate buffer,37°C, 1hr)

Dry, reconstitute iniPrOH

B

A

GP hydrazine (in 70%

70% MeOH 70% MeOH

C

MeOH, 5%HOAc, RT, 14 hr)

70% MeOH, 5% HOAcRecycling

3 x 1 mL MeOH

C18SPE-2MeOH70% → 17%

LC-MSn

Scheme 2. Sample preparation method for analysis of sterols in plasma.

W.J. Griffiths et al. / Free Radical Biology and Medicine 59 (2013) 69–8472

Instrumentation

Sample preparation

The following instruments were used in the sample prepara-tion steps: ultrasonic bath, Grant XB3 (VWR Jencons); multispeedrefrigerated centrifuge, ALC, PK12R; vacuum manifold (AgilentTechnologies); ScanLaf ScanSpeed vacuum concentrator; vortexmixer.

Sample analysis

Chromatographic separation of derivatized sterols was per-formed on either an UltiMate 3000 HPLC system or an UltiMate3000 Binary RSLCnano system, both operated with a conventionalflow rate configuration (both from Dionex, Surrey, UK) utilizing aHypersil GOLD RP column (1.9-mm particles, 50�2.1 mm; FisherScientific, Loughborough, UK). MS analysis was performed withthe chromatographic eluent directed to the atmospheric pressureionization (API) source of an LTQ-Orbitrap XL or LTQ-OrbitrapVelos MS (Thermo Fisher, San Jose, CA, USA). These instrumentshave the hybrid linear ion-trap–Orbitrap analyzer format. TheOrbitrap is a Fourier transform (FT) mass analyzer capable of highresolution (up to 100,000 FWHM) and exact mass measurement(o5 ppm).

Although the work described in this article employed LTQ-Orbitrap instruments the methodology can also be exploited usingtandem-quadrupole, Q-TOF, and ion trap instruments [41,44].

Protocol

Extraction of plasma for analysis of sterols

Plasma (100 ml) was added drop-wise into a 2-ml microcentrifugetube (Eppendorf, Cambridge, UK) containing 1.05 ml of absoluteethanol containing 5 ml of 24(R/S)-[26,26,26,27,27,27-2H6]hydroxy-cholesterol (Avanti Polar Lipids) in propan-2-ol (4 ng/ml) in anultrasonic bath and sonicated for 5 min (empirically we have foundthat drop-wise addition of plasma to ethanol is important to max-imize the extraction of sterols from plasma proteins). The resultantsolution was diluted to 70% (v/v) ethanol by the addition of 0.35 ml ofwater, ultrasonicated for a further 5 min, and centrifuged at 14,000g

at 4 1C for 30 min. If cholesterol or sterols of similar polarity are theobject of the study then 5 ml of [25,26,26,26,27,27,27-2H7]cholesterol(Avanti Polar Lipids) in propan-2-ol (4 mg/ml) is added with 24(R/S)-[26,26,26,27,27,27-2H6]hydroxycholesterol to the ethanol solvent.

A 200-mg Certified Sep-Pak C18 cartridge (Waters) was rinsedwith 4 ml of absolute ethanol followed by 6 ml of 70% ethanol (v/v). Plasma in 70% ethanol (1.5 ml) was applied to the cartridgeand allowed to flow at a rate of �0.25 ml/min (although this flowrate is rather slow, it provides secure extraction of the desiredanalytes on the SPE column). Flow was aided by application of aslight pressure from a BD Biosciences Luer-lock syringe (Sigma–Aldrich). Alternatively, flow was assisted by the use of a vacuummanifold providing a negative pressure at the column outlet. Theflowthrough (1.5 ml) and a column wash of 5.5 ml of 70% ethanolwere collected in a 15-ml Corning centrifuge tube (Fisher Scien-tific) or 12-ml Greiner tube (Sigma–Aldrich) (i.e., SPE-1-Fr-1, 7 ml70% ethanol; Scheme 2). By testing the method with a solution ofcholesterol and 24(R/S)-[26,26,26,27,27,27-2H6]hydroxycholes-terol in 70% ethanol, we found that cholesterol was retainedon the column even after the 5.5-ml column wash, whereas24(R/S)-[26,26,26,27,27,27-2H6]hydroxycholesterol elutes in theflowthrough and column wash. After a further column wash with4 ml of 70% ethanol (SPE-1-Fr-2), cholesterol was eluted from the

column in 2 ml of absolute ethanol (SPE-1-Fr-3). The column wasfurther stripped with an additional 2 ml of absolute ethanol toelute more hydrophobic sterols (SPE-1-Fr-4). Each fraction wasdivided into two subfractions, A and B, and transferred by pipette(using polypropylene tips) into 4-ml polypropylene Cultubes(Simport, Beloeil, QC, Canada) or 12-ml Greiner tubes and driedunder reduced pressure using a vacuum concentrator (Scheme 2).The entire procedure was repeated with an additional 50 or100 ml of plasma, but the final division into A and B subfractionsfollowed by drying down was omitted to leave an intactfraction C.

Enzyme-assisted derivatization

The dried sterol fractions A from above were reconstituted in100 ml of propan-2-ol and vortex mixed thoroughly (2 min), and1 ml of 50 mM phosphate buffer (KH2PO4, pH 7; Sigma–Aldrich)containing 3.0 ml of cholesterol oxidase from Streptomyces

sp. (2 mg/ml in H2O, 44 U/mg protein) was added to each. Aftera further vortex (2 min), the mixture was incubated at 37 1C for1 h and then quenched with 2 ml of methanol. Glacial acetic acid(150 ml) was added to the reaction mixture above (now in �70%methanol, v/v) followed by 150 mg of GP reagent. The mixturewas thoroughly vortexed (2 min) and incubated at room tem-perature overnight in the dark. For the treatment of fractions Bcholesterol oxidase was omitted from the procedure.

SPE of derivatized sterols

Even when derivatized with GP reagent, sterols may bedifficult to solubilize (or retain in solution) when using a highlyaqueous mixture of methanol and water. This can make theextraction of the desired analytes using RP-SPE challenging, asinsoluble material will not be extracted by the stationary phase ofthe column and will be lost from the analysis. To circumvent thisproblem a recycling procedure was used [41,45].

W.J. Griffiths et al. / Free Radical Biology and Medicine 59 (2013) 69–84 73

A 200-mg Certified Sep-Pak C18 cartridge (SPE-2) was washedwith 6 ml of 100% methanol, 6 ml of 10% methanol (v/v), andconditioned with 4 ml of 70% methanol (v/v). The derivatizationmixture from above (�3 ml of 70% methanol, 5% acetic acid, 3%propanol-2-ol, containing 150 mg of GP reagent and 6 mg ofcholesterol oxidase) was applied to the column followed by1 ml of 70% methanol (used to rinse the reaction tube) and 1 mlof 35% methanol. The combined effluent (5 ml) was collected in a15-ml Corning polypropylene tube and diluted with water (4 ml)to give 9 ml of �35% methanol. The resulting solution was againapplied to the column followed by a wash with 1 ml of 17%methanol. The combined effluent (10 ml) was added to 9 ml ofwater in a 50-ml Corning polypropylene tube (or divided into two5-ml aliquots and decanted into two 12-ml Greiner polypropy-lene tubes each containing 4.5 ml of water) to give 19 ml of�17.5% methanol. This solution was again applied to the columnfollowed by a wash with 6 ml of 10% methanol. At this point allthe derivatized sterols were extracted by the column and excessderivatization reagent was in the flowthrough and wash. Deriva-tized sterols were then eluted in three 1-ml portions of 100%methanol and collected in microcentrifuge tubes (SPE-2-Fr-1, -Fr-2, -Fr-3) and combined (either as SPE-2-Fr-1,2 or as SPE-2-Fr-1,2,3); any remaining sterols were eluted with 1 ml of absoluteethanol (SPE-2-Fr-4). LC-MSn analysis revealed that the deriva-tized oxysterols were present in the first 2 ml of methanol eluent(SPE-2-Fr-1, SPE-2-Fr-2), whereas derivatized cholesterol alsotailed into the third milliliter (SPE-2-Fr-3). Acidic sterols elutedpredominantly in the first milliliter of methanol (SPE-2-Fr-1).

To differentiate sterols that naturally possess a 3-oxo functionfrom those oxidized to contain one, samples were analyzed inparallel in the presence (fractions A, e.g., SPE-1-Fr-1A) andabsence (fractions B, e.g., SPE-1-Fr-1B) of cholesterol oxidase(see Scheme 2).

SPE of Underivatized sterols

Sterols in fraction C were extracted and recycled on a secondSPE column essentially as described for the derivatized sterols.An initial solution of 70% ethanol was applied to the column,eluted, and diluted to 23% ethanol/11% methanol/66% H2O (v/v/v).This procedure was repeated to give an 11% ethanol/6% methanol/83% H2O solution at which point all sterols were extracted by thecolumn. After a wash with 10% methanol the sterols were elutedwith four 1-ml portions of methanol and combined.

LC-ESI-MSn on the LTQ-Orbitrap

Chromatographic separation of GP-tagged sterols was per-formed on either an UltiMate 3000 HPLC system or an UltiMate3000 Binary RSLCnano system (both Dionex) utilizing a HypersilGOLD RP column (1.9-mm particles, 50�2.1 mm; Fisher Scienti-fic). Mobile phase A consisted of 33.3% methanol, 16.7% acetoni-trile, containing 0.1% formic acid, and mobile phase B consisted of63.3% methanol, 31.7% acetonitrile containing 0.1% formic acid.After 1 min at 20% B, the proportion of B was raised to 80% B overthe next 7 min and maintained at 80% B for a further 5 min, beforereturning to 20% B in 6 s and reequilibrating for a further 3 min 54s, giving a total run time of 17 min. The flow rate was maintainedat 200 ml/min and the eluent directed to the API source of an LTQ-Orbitrap XL or LTQ-Orbitrap Velos (both Thermo Fisher, San Jose,CA, USA) MS.

The Orbitrap was calibrated externally before each analyticalsession. Mass accuracy was in general better than 5 ppm on theXL and better than 1 ppm on the Velos during the entire analyticalsession. For LC-ESI-MS and LC-ESI-MSn analysis of referencecompounds, sample (1 pg/ml in 60% methanol, 0.1% formic acid)

was injected (20 ml) onto the RP column and eluted into theLTQ-Orbitrap at a flow rate of 200 ml/min. For the analysis ofGP-tagged oxysterols and cholestenoic acids from plasma, 12 ml ofthe combined methanol fractions (2 ml) from the second Sep-PakC18 cartridge (SPE-2-Fr-1,2; equivalent to 0.3 ml of plasma assum-ing all the relevant sterols elute in the first two methanolfractions) was diluted with 8 ml 0.1% formic acid and 20 mlinjected onto the LC column. Two experimental methods wereutilized. In the first experimental method (Experiment 1) threescan events were performed: first an FT-MS scan in the Orbitrapanalyzer over the m/z range 400–650 (or 300–800) at 30,000resolution (FWHM) with a maximum ion fill time of 500 ms,followed by data-dependent MS2 and MS3 events performed inthe LIT with maximum ion fill times of 200 ms on the XL and100 ms on the Velos. For the MS2 and MS3 scans performed on theXL, three microscans were performed, whereas only one micro-scan was required by the Velos; the precursor-ion isolation widthwas set at 2 (to select the monoisotopic ion) and the normalizedcollision energy was at 30 and 35 (instrument settings), respec-tively. A precursor-ion include list was defined according to them/z of the [M]þ ions of predicted sterols so that MS2 waspreferentially performed on these ions in the LIT if their intensityexceeded a preset minimum (500 counts). If a fragment ioncorresponding to a neutral loss of 79 Da from the precursor ionwas observed in the MS2 event and was above a minimal signalsetting (200 counts), MS3 was performed on this fragment(Scheme 3). To maximize efficiency, the MS2 and MS3 eventswere performed in the LIT at the same time that the high-resolution mass spectrum was being recorded in the Orbitrap.The second experimental method (Experiment 2) involved atargeted MRM-like approach. In event 1 the Orbitrap analyzerwas scanned as above, in the second event the MS3 transition, e.g.,534.4-455.4-, was monitored using collision energies of 30 and35 for MS2 and MS3, respectively. In the third event another MS3

transition, e.g., 540.4-461.4-, was monitored in a similarmanner (e.g., to accommodate the 24(R/S)-[2H6]hydroxycholes-terol internal standard). The precursor-ion include list and MRMtransitions utilized for the analysis of plasma are given inSupplementary Table S1. Other fractions from the SPE columnswere analyzed in an identical fashion.

Shotgun ESI-MS on the LTQ-Orbitrap

Underivatized sterols were analyzed by negative-ion ESI-MSon the LTQ-Orbitrap XL. Sample introduction was achieved usingan Advion TriVersa NanoMate (Advion BioSciences, USA) ESI chiputilizing nano-ESI nozzles. Accurate mass spectra (o5 ppm) wererecorded at up to 100,000 (FWHM) resolution over the m/z range100–1000. Peaks appearing at an m/z appropriate to a C24 or C27

bile acid or sulfated and/or glucuronidated sterol were subjectedto MS2 when of appropriate intensity. MS2 spectra were gener-ated in the LIT and recorded using either the LIT or the Orbitrapdetector. To avoid the low-mass cutoff inherent to ion traps, MS2

spectra were also recorded in the ‘‘pulsed Q collision induceddissociation’’ (PQD) mode [48]. In some instances the eluent fromthe second Sep-Pak C18 column was concentrated to 100 ml beforeanalysis to enhance signal intensity.

Results

General considerations related to the EADSA methodology

In this article we describe enzyme-assisted derivatizationwith charge-tagging using the GP hydrazine reagent (Scheme 1).The advantages provided by this form of derivatization include:

Scheme 3. Fragmentation of GP-tagged sterols. (A) MS2 ([M]þ-) fragmentation and (B) MS3 ([M]þ-[M�79]þ-) fragmentation illustrated for GP-tagged

3b-hydroxycholest-(25R)-5-en-26-oic acid. (C) The effect of 7-hydroxylation is illustrated for GP-tagged 3b,7a-dihydroxycholest-(25R)-5-en-26-oic acid.

W.J. Griffiths et al. / Free Radical Biology and Medicine 59 (2013) 69–8474

(i) enhanced lipid solubility in solvents used for RP chromato-graphy; (ii) improved gas-phase ion formation via the ESI process;(iii) derivative-specific MS2 spectra, i.e., prominent loss of 79 Dafrom the molecular ion; and (iv) structurally informative MS3

([Mþ]-[M-79]þ-) spectra (Scheme 3). A library of MS2 and MS3

spectra of authentic standards can be found at http://sterolanalysis.org.uk/. The utilization of the MS3 scan provides an extradimension of separation as only ions fragmenting in the MS2

event by loss of 79 Da are fragmented further in MS3. A drawbackof the charge-tagging methodology is that sterols with a labile7-hydroxy group, e.g., 7-hydroxy-3-oxocholest-4-en-26-oic acids,can dehydrate to a minor extent during sample preparation togive conjugated dienes. This probably accounts for a proportion of3-oxocholesta-4,6-dien-26-oic acid found in plasma (Table S2).A further complication is that 24-oxo-26(or 27) acids eliminateCO2 to give 26-nor-sterols accounting for the presumptive identi-fication of 7a-hydroxy-26-nor-cholest-4-ene-3,24-dione. Similardecarboxylation reactions of 24-oxo-26 acids have been reportedby Yuri et al. [49], Bun-ya et al. [50], and Karlaganis et al. [51]. Thesample preparation method utilized here has been designed withoxysterols and their acidic metabolites in mind; however, it isalso applicable to other steroids possessing an oxo group, e.g.,dehydroepiandrosterone (DHEA), testosterone, or their sulfates.

Identification and quantification

Here we regard a sterol to be ‘‘identified’’ when its retentiontime, exact mass, and MSn spectra are identical to those of theauthentic standard. The term ‘‘presumptively identified’’ is usedwhen the retention time, exact mass, and MSn spectra arecompatible with those of the proposed structure, but there is anabsence of authentic standard. Quantification of monohydroxy-cholesterols was achieved by stable-isotope dilution MS using24(R/S)-[2H6]hydroxycholesterol as the internal standard [45].Quantitative estimates were made for other cholesterol metabo-lites; this is possible using 24(R/S)-[2H6]hydroxycholesterol as theinternal standard, as previous studies have shown that GP-tagged3-oxo-4-ene sterols give a similar mass spectrometric responseirrespective of the other functional groups attached to the steroidskeleton [43]. For accurate quantification calibration curvesshould be generated using authentic standards. In this study thelevels of free sterols are measured. Sterols may also be esterified,glucuronidated, and/or sulfated. The last two conjugates areinvestigated here using shotgun ESI-MS but are not quantified.If sterols esterified to fatty acids are the subject of the study theintact molecules may be analyzed directly by LC-ESI-MS afterappropriate extraction and purification [25]. Alternatively, the

W.J. Griffiths et al. / Free Radical Biology and Medicine 59 (2013) 69–84 75

free sterols can be analyzed following saponification and EADSA[52].

Whereas fractions SPE-2-Fr1,2A in the sample preparationprotocol will contain the GP charge-tagged stable-isotope-labeledstandard 24(R/S)-[2H6]hydroxycholesterol, the absence of choles-terol oxidase in the parallel procedure generating SPE-2-Fr-1,2Bmeans that 24(R/S)-[2H6]hydroxycholesterol in this fraction is not

derivatized and thus does not act as an internal standard forquantification. One solution to this problem is to generate, e.g., a22R-[2H7]hydroxycholest-4-en-3-one reference standard from22R-[2H7]hydroxycholesterol (Avanti Polar Lipids). The isotope-labeled 3-oxo-4-ene compound can then be added to ethanolduring the initial extraction step and can be used for quantifica-tion of sterols in SPE-2-Fr-1,2B. Alternatively, DHEA 3-sulfate,which is abundant in plasma and possesses a 17-oxo group and isnot oxidized further by cholesterol oxidase, will appear in bothSPE-2-Fr-1,2A and SPE-2-Fr-1,2B fractions and can be used tonormalize fraction B to fraction A.

Oxysterols

(25R),26-Hydroxycholesterol is the most abundant oxysterolpresent in adult plasma (Fig. 1). The level of free (25R),26-hydroxycholesterol determined here (19.1270.70 ng/ml, meanof 84 adult subjects7standard error) is in good agreement withvalues of the free oxysterol determined by GC–MS (13–41 ng/ml),which tend to be about 10% of the combined free and fattyacid ester values [22,53]. In addition to (25R),26-hydroxycholes-terol, 7a- (0.8270.39 ng/ml), 24S- (6.8670.31 ng/ml), and

80

90

100

40

50

60

70

ng/m

L

10

20

30

40

n-24

-oic

aci

d

n-24

-oic

aci

dn-

24-o

ic a

cid

-5-e

n-24

-one

-die

ne-3

-one

e-3,

24-d

ione

-5-e

n-24

-one

e-3β

,24S

-dio

l

ne-3

β,25

-dio

l

ne-3

β,26

-dio

l

st-4

-en-

3-on

e3β

,7β-

diol

t-5-e

n-7-

one

st-4

-en-

3-on

e

-Hyd

roxy

chol

-5-e

n

oxy-

3-ox

ocho

l-4-e

nih

ydro

xych

ol-5

-e

-Hyd

roxy

chol

est

roxy

chol

esta

-4,6

-

6-no

rcho

lest

-4-e

neox

y-26

-nor

chol

est

Cho

lest

-5-e

ne

Cho

lest

-5-e

n

Cho

lest

-(25R

)-5-e

n

7β-H

ydro

xych

oles

Cho

lest

-5-e

ne-3

3β-H

ydro

xych

oles

7α-H

ydro

xych

oles

3β

7α-H

ydr o

3β,7

α-D 3 β

12α-

Hyd

7α-H

ydro

xy-2

3β,7

α-D

ihyd

ro C 7 3 7

Fig. 1. Levels (ng/ml7SE) of free oxysterols and cholenoic and cholestenoic acids measu

3-oxo-4-ene structure and red bars to molecules with a 3b-hydroxy-5-ene function. Sy

S2. Measured values are given in Supplementary Table S2, as are activities as nuclear

25-hydroxycholesterols (4.0670.22 ng/ml) were found in adultplasma (Fig. 2, Table S2). The levels of free 7a-hydroxycholesteroldetermined here are somewhat lower than values determined byGC–MS (7–15 ng/ml), which correspond to about 10–20% ofthe combined free and fatty acid ester values [22,25,27,53].The downstream metabolite 7a-hydroxycholest-4-en-3-one(2.2970.25 ng/ml) is also found in plasma [53,54].

It should be noted that the methodology described here isprimarily targeted toward side-chain-oxidized sterols (oxysterolsand acids) and there may be underestimation of the levels of both7a-hydroxycholesterol and 7a-hydroxycholest-4-en-3-one, whichmay not be completely recovered in Fr-1 from SPE-1. If theseoxysterols are the target of study SPE-1 may be eluted with astronger solvent.

With the current method we have identified 25-hydroxycho-lesterol and its hydroxylated (7a,25-dihydroxycholesterol,o0.2 ng/ml) and oxidized products (7a,25-dihydroxycholest-4-en-3-one, 1.2470.20 ng/ml) (see Fig. 3, Table S2). AlthoughGC–MS and LC-MS/MS analysis of plasma regularly identifies25-hydroxycholesterol, downstream metabolites are not usuallydetected from healthy subjects because of their low abundance[22,25,27]. Other oxysterols identified in adult plasma include(25R),7a,26-dihydroxycholesterol (1.6570.39 ng/ml) and 7a,26-dihydroxycholest-(25R)-4-en-3-one (5.5570.41 ng/ml). Again,these metabolites are not usually investigated in GC–MS orLC-MS/MS analysis of plasma from healthy subjects [27,30,53].

24S,25-Epoxycholesterol is an unusual oxysterol in that it isformed in parallel to cholesterol via a shunt of the mevalonatepathway by the same enzymes that synthesize cholesterol

3β,7

α-di

ol

ene-

3β,6

-dio

l

6-oi

c ac

id-2

6-oi

c ac

id

st-5

-en-

y-on

e

n-26

-oic

aci

d

st-4

-en-

3-on

eα,

25-tr

iol

R)-4

-en-

3-on

e,2

6-tri

ol

n-26

-oic

aci

d

n-26

-oic

aci

d

n-26

-oic

aci

d

26-o

ic a

cid

or

5-en

-z-o

ne

n-26

-oic

aci

d26

-oic

aci

d

Cho

lest

-5-e

ne-3

Cho

lest

-4-e

hole

sta-

4,6-

dien

-26

ycho

lest

a-5,

7-di

en

x-D

ihyd

roxy

chol

es

chol

est-(

25R

)-5-e

n

5-D

ihyd

roxy

chol

esC

hole

st-5

-ene

-3β,

7

drox

ycho

lest

-(25R

(25R

)-5-e

ne-3

β,7α

ydro

xych

oles

t-5-e

ydro

xych

oles

t-5-e

chol

est-(

25R

)-5-e

n

roxy

chol

est-5

-en-

2rih

ydro

xych

oles

t-5

chol

est-(

25R

)-4-e

nch

oles

t-(25

R)-5

-en-

3-O

xoch

3β-H

ydro

x

3β,x

3β-H

ydro

xyc

7α,2

5 C

7α,2

6-D

ihyd

Cho

lest

-

3β,x

-Dih

y

3β,y

-Dih

y

3β,7

β-D

ihyd

roxy

c

3β,z

-Dih

ydr

3β,x

,y-T

r

7α-H

ydro

xy-3

-oxo

c3β

,7α-

Dih

ydro

xyc

7 3

red in adult plasma (n¼84). Blue bars correspond to endogenous molecules with a

stematic names are used. To translate to common names see Supplementary Table

receptor (e.g., LXR, FXR) agonists.

100

8.27RIC: 534.4054 ± 10 ppm

20.6210.07

%RA

10.539.9320.62 ng/mL

7.92

7.70

10.94

11.088.47

0100

26.72 ng/mL

%RA

00 82 4 6 10 12 14

437

OHMS3: 534→455→7.70 min

100

NHN

C34H52N3O2+

Exact Mass: 534.4054

O

N

*b1-12

*b3-C2H4

%RA

‘*f

151 394

427

353

163*b2

409‘*e

150 200 250 300 350 400 450

3811770

325

m/z

437

MS3: 534→455→7.92 min

OH100

N

HNN

C34H52N3O2+

Exact Mass: 534.4054

O

N

*b3-C2H4

%RA

151394

427

*b1-12

163409

‘*h

150 200 250 300 350 400 450

381

0353

m/z

427

OHMS3: 534→455→8.27 min

100

NHN

C34H52N3O2+

Exact Mass: 534.4054

*b3-C2H4

163

151 437

ON*b1-12

%RA

394

*b2412

150 200 250 300 350 400 450

3831770

353325

m/z

437

MS3: 534→455→10.53 min

100

NHN

N

OH

427

O

C34H52N3O2+Exact Mass: 534.4054

%RA

151

394*b1-12

*b3-C2H4

150 200 250 300 350 400 450

151383

179

0323

409231203

m/z

Fig. 2. (A) LC-MS reconstructed ion chromatograms (RICs) of m/z 534.4054710 ppm corresponding to monohydroxycholesterols modified by EADSA from representative

adult (top) and cerebrotendinous xanthomatosis (CTX) (bottom) plasma samples. The CTX patient was on treatment with chenodeoxycholic acid. The samples were

oxidized with cholesterol oxidase before GP derivatization to convert 3b-hydroxy-5-ene groups to their 3-oxo-4-ene equivalents. The MS3 (534.4-455.4-) spectra of

components eluting at (B) 7.70 min (24S-hydroxycholesterol), (C) 7.92 min (25-hydroxycholesterol), (D) 8.27 min ((25R),26-hydroxycholesterol), and (E) 10.53 min (7a-

hydroxycholesterol and 7a-hydroxycholest-4-en-3-one) are shown. Syn and anti conformers of (25R),26-hydroxycholesterol are observed eluting at 8.27 and 8.47 min, and

those of 7a-hydroxycholesterol at 10.53 and 11.08 min. Other peaks in (A) correspond to 7b-hydroxycholesterol, 7-oxocholesterol, and cholest-4-en-3b,6-diol eluting at

9.93, 10.07, and 10.94 min, respectively. It is likely that cholest-4-en-3b,6-diol is derived from 5,6-epoxycholesterol during the derivatization process. It is probable that

the concentration of 25-hydroxycholesterol is overestimated as there is tailing of the 24S-hydroxycholesterol peak into that of the 25-hydroxy isomer. Concentrations of

the most abundant metabolites are indicated in the chromatograms. Spectra were recorded on the LTQ-Orbitrap XL. %RA, percentage relative abundance.

W.J. Griffiths et al. / Free Radical Biology and Medicine 59 (2013) 69–8476

100

RIC: 550.4003 ± 10 ppm

6.488.14 ng/mL

7.35

6.78

%RA5.97

0100

9.2094.93 ng/mL

%RA9.42

00 82 4 6 10 12 14

453

MS3: 550→471→5.97 min

OH100

NHN

N

OHC34H52N3O3

+

O

Exact Mass: 550.4003%RA

151435

*b1-12*b3-C2H4

179

150 200 250 300 350 400 450

392

0425 443

m/z

MS3: 550→471→6.48 min

MS3: 550→471→9.20 min

453

OH

100

NHN

OH443

O

N Exact Mass: 550.4003Exact Mass: 550.4003

*b1-12

%RA

410

151

392435

*b3-C2H4

179

425

231

150 200 250 300 350 400 4500

381

m/z

C34H52N3O3+ C34H52N3O3

+

453OH100

392N

HNN

OH

435O

%RA

443

151

*b1-12*b3-C2H4 425

407279363323

150 200 250 300 350 400 4500

179 381209

m/z

Fig. 3. (A) LC-MS RICs of m/z 550.4003710 ppm corresponding to GP-tagged dihydroxycholesterols from representative adult (top) and CTX (bottom) plasma samples.

The samples were treated with cholesterol oxidase before GP derivatization to convert 3b-hydroxy-5-ene groups to their 3-oxo-4-ene equivalents. The MS3 (550.4-

471.4-) spectra of components eluting at (B) 5.97 min (7a,25-dihydroxycholesterol and 7a,25-dihydroxycholest-4-en-3-one), (C) 6.48 min ((25R),7a,26-dihydroxycho-

lesterol and 7a,26-dihydroxycholest-(25R)-4-en-3-one)), and (D) 9.20 min (7a,12a-dihydroxycholesterol and 7a,12a-dihydroxycholest-4-en-3-one) are shown. 7a,25-

Dihydroxycholesterol/7a,25-dihydroxycholest-4-en-3-one and (25R),7a,26-dihydroxycholesterol/7a,26-dihydroxycholest-(25R)-4-en-3-one and 7a,12a-dihydroxycho-

lesterol/7a,12a-dihydroxycholest-4-en-3-one elute as syn and anti conformers at 5.97 and 6.78 min, 6.48 and 7.35 min, and 9.20 and 9.42 min, respectively. Spectra in

(B) and (C) are from a healthy adult, (D) is from a CTX patient on treatment with CDCA. Concentrations of the most abundant metabolites are indicated in the

chromatograms. Spectra were recorded on the LTQ-Orbitrap XL.

W.J. Griffiths et al. / Free Radical Biology and Medicine 59 (2013) 69–84 77

[55,56]. 24S,25-Epoxycholesterol has been found previously inplasma at low levels (2 ng/ml) by LC-MS/MS [27]. During oursample preparation we find that 24S,25-epoxycholesterol canisomerize to 24-oxocholesterol [56]. Here we identify 24-oxocho-lesterol at a level of o1 ng/ml (0.5970.85 ng/ml) in plasma,presumably formed from the 24S,25-epoxide.

Cholenoic and cholestenoic acids

The acidic cholesterol metabolites identified in plasma with a3b-hydroxy-5-ene or 3-oxo-4-ene structure include unsaturatedC24 precursors of primary bile acids, i.e., 3b-hydroxychol-5-en-24-oic acid (1.7870.25 ng/ml), 3b,7a-dihydroxychol-5-en-24-oic (4.1370.77 ng/ml), and 7a-hydroxy-3-oxochol-4-en-24-oic(1.9170.26 ng/ml) acids (see Fig. 1). 3b,7a-Dihydroxychol-5-en-24-oic and 7a-hydroxy-3-oxochol-4-en-24-oic acids arepotential precursors of chenodeoxycholic acid in bile acidbiosynthetic pathways [13,57]. The most abundant members ofthe bile acid biosynthesis pathways [13] found in plasma are 3b-hydroxycholest-(25R)-5-en-26-oic (82.6073.50 ng/ml), 3b,7a-dihydroxycholest-(25R)-5-en-26-oic (47.4276.61 ng/ml), and

7a-hydroxy-3-oxocholest-(25R)-4-en-26-oic (63.8975.44 ng/ml; Fig. 4 and 5) acids; these data are consistent with findingsby others using GC–MS [31,53]. Surprisingly, 3b,7b-dihydrox-ycholest-(25R)-5-en-26-oic acid (5.3670.81 ng/ml) was detected inplasma, probably formed by the hepatic mitochondrial epimerizationreaction suggested by Shoda et al. [58].

In addition to 3b,7a- and 3b,7b-dihydroxycholest-(25R)-5-en-26-oic acids in plasma (Fig. 5), two other dihydroxycholest-5-en-26-oic acids were presumptively identified. Their MS3 spectrasuggest the locations of the additional hydroxyl groups to be onthe C-17 side chain. Tentative suggestions for the locations of theadditional hydroxyl groups are C-25 (5.2770.74 ng/ml, Fig. 5B)and C-22 (2.2770.34 ng/ml, Fig. 5C). A third compound gave anMS3 spectrum that could correspond to a dihydroxycholest-5-en-26-oic acid (9.6371.46 ng/ml) or alternatively a trihydroxychol-est-5-enone, possibly 3b,22,25-trihydroxycholest-5-en-24-one(Fig. 5E). Whereas the major monohydroxycholestenoic acid inadult plasma is 3b-hydroxycholest-(25R)-5-en-26-oic acid(82.6073.50 ng/ml), a minor earlier eluting peak in the RIC ofm/z 548.3847 suggested the possible presence of the 25S isomer(Fig. 4A). However, comparison of retention time and MSn spectra

100 7.98RIC: 548.3847 ± 5 ppm

156.95 ng/mL

%RA7.62

0100 7.62

28.46 ng/mL

%RA

00 82 4 6 10 12 14

423OOH

MS3: 548→469→7.62 min

441*b3-C2H4

100

NHN

C34H50N3O3+

Exact Mass: 548.3847*b1-12

163

151O

N%RA

325

380395

253

282

150 200 250 300 350 400 4500

m/z

441O

MS3: 548→ → 469→7.98 min

423

OH100

NHN

C34H50N3O3+

Exact Mass: 548.3847

*b3-C2H4

163

451

151O

N*b1-12%RA

151

395

408*b2

150 200 250 300 350 400 450

3800

325 353177

m/z

Fig. 4. LC-MS RICs of m/z 548.384775 ppm corresponding to 3b,x-dihydroxycholest-5-en-y-one and 3b-hydroxycholest-(25R)-5-en-26-oic acid after EADSA modification

and found in plasma from a representative adult (top) and a CTX patient (bottom). The CTX patient was on treatment with CDCA. Chromatograms are normalized to the

most abundant peak. The samples were oxidized with cholesterol oxidase before GP derivatization to convert 3b-hydroxy-5-ene groups to their 3-oxo-4-ene equivalents.

The MS3 (548.4-469.3 -) spectra of components eluting at (B) 7.62 min (3b,x-dihydroxycholest-5-en-y-one) and (C) 7.98 min (3b-hydroxycholest-(25R)-5-en-26-oic

acid) from adult plasma are shown. The peak at 7.62 min in the chromatogram from the CTX patient gave a spectrum essentially identical to that in (B). Concentrations of

the most abundant metabolites are indicated in the chromatograms. Spectra were recorded on the LTQ-Orbitrap XL.

W.J. Griffiths et al. / Free Radical Biology and Medicine 59 (2013) 69–8478

with those of the authentic standard ruled out this possibility.A possible alternative structure is 3b,22-dihydroxycholest-5-en-24-one (5.6470.50 ng/ml, Fig. 4B).

CTX

CTX is characterized by the absence of a functional CYP27A1enzyme, which precludes the formation of (25R),26-hydroxycho-lesterol and subsequent cholesten-(25R)-26-oic acids [59–61].A consequence of this is that hydroxylated cholesterol metabolitesbecome shunted to alternate metabolic pathways. CTX in earlyinfancy can present with cholestatic liver disease, in early child-hood with chronic diarrhea and cataracts, in later childhood withtendon xanthomata and learning difficulties, and in adult life withspastic paraparesis, with a fall in IQ or dementia, with ataxia and/or dysarthria, with seizures, or with peripheral neuropathy [61].Treatment is with CDCA. In Figs. 2–5 representative data areshown for CTX patients on treatment with simvastatin and CDCAat the time of sampling. From profiling the oxysterol content ofplasma of CTX patients the following diagnostic features weredefined: (1) an absence of unsaturated C24 acids; (2) a majorreduction in the level of (25R),26-hydroxycholesterol (Fig. 2A); (3)

elevation in the level of 7a-hydroxycholest-4-en-3-one; (4) anabsence of 3b-hydroxycholest-(25R)-5-en-26-oic, 3b,7a-dihydrox-ycholest-(25R)-5-en-26-oic, and 7a-hydroxy-3-oxocholest-(25R)-4-en-26-oic acids (Fig. 4A and 5A); and (5) a great enhancement inthe level of 7a,12a-dihydroxycholest-4-en-3-one (Fig. 3A). Thesefeatures were independent of treatment with CDCA and simvas-tatin and are also observed for patients not on treatment. How-ever, the levels of 7a-hydroxycholest-4-en-3-one are lower inpatients treated with CDCA than in untreated patients. Thesediagnostic features, summarized in Table 2, are revealed fromthe analysis of microliter quantities of plasma (easily obtainablefrom infants) and in combination are unique to CTX. The absenceof a functional CYP27A1 enzyme is directly revealed here by anabsence of (or great reduction in) its direct metabolic products.This may be seen as a diagnostic advantage over alternativeindirect tests based on an elevated abundance of metabolitesshunted into other bile acid/alcohol biosynthetic pathways.

Shotgun analysis

Shotgun ESI-MS in the negative-ion mode is appropriate forthe analysis of acidic cholesterol metabolites, particularly sulfates

100

RIC: 564.3796 ± 10 ppm

105.97 ng/mL6.23

%RA2.76

4.56

5.463.72

7.00

0100

5.11 ng/mL

%RA

00 82 4 6 10 12 14

467O

MS3: 564→485→2.76 min

OHOH

100

NC34H50N3O4

+

Exact Mass: 564.3796HN

O

N%RA

*b1-12

*b3-C2H4

163

150 200 250 300 350 400 450

151 421439

253457

0

163325

m/z

457OOH

MS3: 564→485→3.72 min

439

OH

*b3-C2H4

100

NC34H50N3O4

+

Exact Mass: 564.3796163

467

HN

O

N%RA

325

253

467

*b1-12 369 396

150 200 250 300 350 400 450

151253

0

m/z

467

OMS3: 564→485→4.56 min

457

467OH100

NC H N O +

OH

151

HN

O

NC34H50N3O4

+

Exact Mass: 564.3796*b1-12%RA

151

439*b3-C2H4

179

421

150 200 250 300 350 400 4500

179

393

m/z

OH O100

MS3: 564→485→5.46 min 385

OH

N

HN

O

N

C34H50N3O4+

Exact Mass: 564 3796Exact Mass: 564.3796%RA

467

*b1-12

*b3-C2H4

163 423

150 200 250 300 350 400 450

151 439457

467

0

163 423367325251

m/z

467O

MS3: 564→485→6.23 min

OH100

NC O +

OH

457

HN

O

NC34H50N3O4

+

Exact Mass: 564.3796

*b1-12

%RA

151 439

*b3-C2H4424

449421

395

150 200 250 300 350 400 4500

179 406231

253378

m/z

Fig. 5. (A) LC-MS RICs of m/z 564.3796710 ppm corresponding to dihydroxycholestenoic acids and trihydroxycholestenones after EADSA modification from

representative adult (top) and CTX (bottom) plasma samples. The CTX patient was on treatment with CDCA. The samples were oxidized with cholesterol oxidase before

GP derivatization to convert 3b-hydroxy-5-ene groups to their 3-oxo-4-ene equivalents. The MS3 (564.4-485.3-) spectra of components eluting at (B) 2.76 min (3b,

x-dihydroxycholest-5-en-26-oic acid), (C) 3.72 min (3b,y-dihydroxycholest-5-en-26-oic acid), (D) 4.56 min (3b,7b-dihydroxycholest-(25R)-5-en-26-oic acid), (E) 5.46 min

(3b,z-dihydroxycholest-5-en-26-oic acid or 3b,x,y-trihydroxychelest-5-en-z-one), and (F) 6.23 min (3b,7a-dihydroxycholest-(25R)-5-en-26-oic and 7a-hydroxy-3-

oxocholest-(25R)-4-en-26-oic acid) are shown. 3b,7a-Dihydroxycholest-(25R)-5-en-26-oic/7a-hydroxy-3-oxocholest-(25R)-4-en-26-oic acid appear as syn and anti

conformers (6.23 and 7.00 min). Spectra in (B–F) were from a healthy adult. Where spectra match those of authentic standards chemical structures are given in the

legend. Locations x, y, and z are probably on the side chain of the sterol. Suggested structures of the GP-tagged molecules are shown. Concentrations of the most abundant

metabolites are indicated in the chromatograms. Spectra were recorded on the LTQ-Orbitrap XL.

W.J. Griffiths et al. / Free Radical Biology and Medicine 59 (2013) 69–84 79

Table 2Most important diagnostic features of CTX.

EADSA (plasma) Shotgun ESI-MS (plasma) Alternative (ESI-MS urine)

Absence of (or major reduction in) Abundant glucuronides of Abundant glucuronides of

� (25R),26-Hydroxycholesterol

� 3b-Hydroxycholest-(25R)-5-en-26-oic acid

� 3b,7a-Dihydroxycholest-(25R)-5-en-26-oic acid

� 7a-Hydroxy-3-oxocholest-(25R)-4-en-26-oic acid

� Cholestanetetrols

� Cholestanepentols

� Cholestanehexols

� Cholestanetetrols

� Cholestanepentols

� Cholestanehexols

Elevation in

� 7a-Hydroxycholest-4-en-3-one

� 7a,12a-Dihydroxycholest-4-en-3-one

W.J. Griffiths et al. / Free Radical Biology and Medicine 59 (2013) 69–8480

and sulfonates. The method is fast and simple and, whenperformed at high resolution with exact mass measurements,provides elemental formula information. When complemented byMS2, structural information is also forthcoming; however, theabsence of online chromatography means that structural isomersare not differentiated. Shotgun ESI-MS is highly complementaryto EADSA, as it is appropriate for sulfate esters, many of which arelocated at C-3 and thus invisible to EADSA methodology. In termsof clinical chemistry, shotgun ESI-MS offers two importantadvantages over LC-MS-based EADSA: (1) it is high-throughput,with MS and MS2 spectra being recorded in less than a minute,and (2) by utilizing an individual nano-ESI nozzle for each sample,problems of ‘‘carryover’’ are eliminated. Alternatively, a negative-ion LC-ESI-MS approach could be adopted, in which case carry-over between samples can be a problem, and the penalty ofchromatographically resolving isomers is paid in analysis time.Further, authentic standards for sterol glucuronides and sulfates,other than for cholesterol sulfate, are not commercially available,making isomer identification essentially impossible.

Adults and infants

Shotgun analysis was performed on the plasma extract deli-pidated on SPE-1 and desalted on SPE-2. No further chromato-graphy was performed, hence the resulting spectra werecomplicated by numerous endogenous molecules other thansterols, including free fatty acids, which tend to dominate thenegative-ion spectra, e.g., palmitic acid (m/z 255.2334) and stearicacid (m/z 283.2645), and exogenous molecules such as plastici-zers from polypropylene storage vessels (Fig. 6A). Despite this,high-resolution accurate mass spectra of healthy infants andadult plasma samples show the presence of sulfates of C19 andC21 steroids, particularly DHEA (m/z 367.1590); isomers of hydro-xyandrostanone, e.g., androsterone, epiandrosterone, and andros-tenediol (m/z 369.1761); pregnenediol (m/z 397.2029); andpregnenetriol (m/z 413.2005) (Fig. 6A). MS2 performed in theLIT exploiting PQD allows the confirmation of the presence of thesulfate group by the characteristic fragment ion at 97 (HSO4

�)(Fig. 6C).

Shotgun analysis of CTX plasma

In comparison to normal plasma, the untreated CTX patientgave a spectrum dramatically different at the high m/z end (m/z600–650), where peaks characteristic of glucuronides of choles-tanetetrols (m/z 611.3798), pentols (m/z 627.3737), and hexols(m/z 643.3680) are evident (cf. Fig. 6A and B) [60,62]. Whereasaccurate mass measurements (o5 ppm) made at high resolutionsuggest an elemental composition, further structural informationis forthcoming by performing MS2. This is evident in the MS2

spectra of the three glucuronides, which show neutral losses of

60, 118, 176, and 194 characteristic of the glucuronic acid group[63] (Fig. 6D). From the MS2 spectra we are unable to define thestereochemistry of the cholestane skeleton; however, detailedstudies on glucuronides found in the bile of a CTX patient indicatea predominant 5b-cholestane-3a,7a,12a,25-tetrol skeleton withglucuronidation at the 3a position [64]. Alternative and additionalsites of side-chain hydroxylation are at C-22, C-23, and C-24 [60].In patients on CDCA treatment, [M�H]� ions corresponding tothese glucuronides are absent from the shotgun spectra.

Shotgun analysis of plasma from a patient with suspected

hydroxysteroid dehydrogenase 3B7 deficiency

Hydroxysteroid dehydrogenase (HSD) 3B7 (also known as3b-hydroxy-C27-steroid oxidoreductase or 3b-hydroxysteroid-D5-C27-steroid dehydrogenase) is the enzyme responsible for convert-ing C27 sterols with a 3b,7a-dihydroxy-5-ene structure to those witha 7a-hydroxy-3-oxo-4-ene structure in the bile acid biosynthesispathway. An absence of functional enzyme results in a buildup ofsterols and bile acids retaining the 3b-hydroxy-5-ene function [61].In the circulation these are found sulfated and/or glucuronidated.Shown in Fig. 7 is the shotgun ESI-MS spectrum of a child withsuspected HSD3B7 deficiency. It is likely that sulfation is predomi-nantly at position C-3, in which case many of the metabolitesobserved in this spectrum will be transparent to EADSA. Thus, in thiscase shotgun ESI-MS provides invaluable information that would belost by simply performing EADSA. HSD3B7 deficiency may presentwith neonatal conjugated hyperbilirubinemia, rickets, hepatome-galy, pruritus, and steatorrhea. Provided that liver damage is not tooadvanced at the time of diagnosis, the condition responds extremelywell to bile acid replacement therapy [61].

Caveats

In this article we have extolled the virtues of EADSA incombination with shotgun ESI-MS for the analysis of cholesterolmetabolites. The two methods are complementary, as EADSAworks with steroids possessing a free 3b-hydroxy or oxo groupand shotgun negative-ion ESI-MS works most effectively forsteroids conjugated with a strongly acidic group, e.g., sulfuricacid ester, in which the site of modification is often at C-3, makingthe molecule transparent to EADSA, but readily ionized bynegative-ion ESI.

To the best of our knowledge, EADSA offers sensitivity advan-tages for the analysis of sterols that exceed those of any othermethodology. There are, however, a number of disadvantages ofEADSA methodology, which should be considered.

1.

Steroids with a free 3a-hydroxy group are not oxidized bycholesterol oxidase [47]; thus, e.g., primary bile acids cannot

●

100255.2334

283 .2645●

%RA311.1691

367.1590473 2831

*

397.2029

413 2005

*

.565.3700

*

250 300 350 400 450 500 5500

.

600

** *

m/z

283 2646●

100

255.2334

.

●

295.2284473.2828

%RA

367.1590539.5060* 611.3798

■

397.2052*

627.3737■

■

250 300 350 400 450 500 5500

600*

643.3680

m/z

[M-H]-367

MS2: 367→ MS2: 611→

100O

O%RASO OO

C19H27O5S- C33H55O10-Exact Mass: 367.1585

HSO4-97

50 100 200 300 4000

m/z

593100

OH

O OHOH

HOO

HO OH

CO2

%RAH

Exact Mass: 611.3801

551

[M-H-176]-

493

200 300 400 500 6000

417435

m/z

Fig. 6. Negative-ion ESI-MS spectra of plasma samples from (A) a control infant and (B) a boy with CTX not on treatment. (C) ESI-MS2 spectrum of the ion m/z 367 from the

control sample. (D) ESI-MS2 spectrum of the ion m/z 611 from the boy with CTX. Peaks labeled with an asterisk correspond to C19 and C21 steroid sulfates, those with a

filled square to oxysterol glucuronides, and those with a filled circle to fatty acids. Spectra were recorded on the LTQ-Orbitrap XL. Spectra in (A) and (B) were recorded at

high resolution, (C) and (D) utilized the ion-trap detector.

W.J. Griffiths et al. / Free Radical Biology and Medicine 59 (2013) 69–84 81

be converted to 3-ketones and subsequently derivatized.However, this problem can be circumvented by the use of3a-hydroxysteroid dehydrogenase instead of cholesterol oxi-dase, which generates the 3-ketone suitable for subsequentderivatization.

2.

A problem with all steroid derivatization reactions is theprevalence of labile hydroxy groups to be eliminated as waterduring the derivatization process. This is seen with the methodpresented here for 7a-hydroxy-3-oxocholest-(25R)-4-en-26-oic acid, which undergoes dehydration to the 4,6-diene despitethe reaction being performed at room temperature.

3.

Steroids possessing an epoxide group are labile in acidic solventsand we find that 24S,25-epoxycholesterol becomes isomerized tothe 24-ketone, hydrolyzed to the 24,25-diol, and methanolyzedto the 24-hydroxy-25-methyl ether (and/or 25-hydroxy-24-methyl ether) during the EADSA process. 5,6-Epoxycholesterolbecomes hydrolyzed to the 3b,5,6-triol, which eliminates waterto leave the cholest-4-ene-3b,6-diol [56].

4.

A disadvantage of the hydrazone derivative is that its forma-tion is reversible in acidic solvents. This precludes the storageof derivatized steroids in acidic solvents and dictates thatGP-derivatized steroids are not kept for long periods (424 h)in their injection solvent before injection onto the LC system.However, we find that when stored in 100% methanol GP-derivatized steroids are stable.

5.

A consequence of derivatization with a hydrazine reagent isthe formation of syn and anti hydrazones. These are often

resolved in our chromatography system, highlighting thequality of the chromatography, but complicating the ultimatechromatogram.

6.

The chromatographic method described here utilizes a Hyper-sil Gold RP column. By employing the sample preparationprocedure described here columns have an exceedingly longlifetime (all columns purchased in the past 4 years are still inuse today with no reduction in performance). However, thereis some batch-to-batch variation, which means that with somecolumns 25-hydroxycholesterol is only partially resolved fromthe 24S-isomer. This problem can be overcome by lengtheningthe gradient but at the expense of throughput.

7.

In the EADSA methodology reported here quantification ismade using 24(R/S)-[2H6]hydroxycholesterol as the internalstandard. Cholesterol oxidase from Streptomyces sp. has similaractivities toward most sterols with a 3b-hydroxy-5-ene struc-ture [47], and we have found that 3-oxo-4-ene sterols oncederivatized with GP-hydrazine give essentially the same ESIresponse irrespective of other functional groups present [43],allowing good quantitative estimates to be made without theinclusion of additional reference standards other than[2H7]cholesterol for sterols eluting in Fr-3 from SPE-1. How-ever, the sample preparation protocol described here has beendesigned with side-chain-oxidized sterols and cholestenoicacids in mind, and it is likely that 7a-hydroxycholesterol andother ring-oxidized sterols are not fully eluted from SPE-1 inFr-1; hence their amount is probably underestimated. This is

Fig. 7. (A) Negative-ion ESI-MS spectrum of a plasma sample from an infant with suspected HSD3B7 deficiency. The inset shows the high m/z range 640–700. MS2 spectra

ions of (B) m/z 280, (C) m/z 328 and 657, and (D) m/z 336 and 673, corresponding to cholest-5-ene-3b,x-diol disulfate [M�2H]2� , cholest-5-ene-3b,x-diol 3-sulfate and

x-glucuronide [M�2H]2� and [M�H]� , and cholest-5-ene-3b,x,y-triol 3-sulfate and x-glucuronide [M�2H]2� and [M�H]� ions, respectively, are shown. Sulfation is

assumed at C-3, and x and y are drawn on the side chain. Peaks labeled with an asterisk correspond to oxysterol mono- or disulfates, those with a filled triangle to

oxysterols sulfated and glucuronidated, that with a diamond to taurine-conjugated bile acid sulfates, and that with a filled circle to fatty acids. Spectra were recorded on

the LTQ-Orbitrap XL and at high resolution.

W.J. Griffiths et al. / Free Radical Biology and Medicine 59 (2013) 69–8482

easily rectified by using a stronger eluent to elute 7a-hydro-xycholesterol from SPE-1.

8.

A number of sterols can exist naturally in their 3-oxo-4-eneand 3b-hydroxy-5-ene forms; this dictates the analysis of A(with cholesterol oxidase) and B (without cholesterol oxidase)fractions, and the difference A � B equals the amount of free3b-hydroxy-5-ene metabolite. In most studies we use DHEA3-sulfate, an abundant steroid present in plasma, not oxidizedby cholesterol oxidase, as a natural internal standard to allowthe normalization of fraction B to the fraction A equivalents.Alternatively, 22R-[2H7]hydroxycholest-4-en-3-one can beused as an internal standard for B fractions.

Despite these caveats, the advantages of EADSA lie in itssimplicity and robustness, offering greatly improved sensitivityfor the analysis of steroids and sterols.

Acknowledgments

Work in Swansea was supported by funding from the UKResearch Councils BBSRC (BBC5157712, BBC5113561, BBI0017351to W.J.G., BBH0010181 to Y.W.) and EPSRC (studentship to M.O.).The EPSRC National Mass Spectrometry Service Centre is warmly

acknowledged for providing access to the LTQ-Orbitrap XL massspectrometer. Members of the European Network for OxysterolResearch are thanked for informative discussions.

Appendix A. Supporting information

Supplementary data associated with this article can be found inthe online version at http://dx.doi.org/10.1016/j.freeradbiomed.2012.07.027.

References

[1] Bauman, D. R.; Bitmansour, A. D.; McDonald, J. G.; Thompson, B. M.; Liang, G.;Russell, D. W. 25-Hydroxycholesterol secreted by macrophages in response toToll-like receptor activation suppresses immunoglobulin A production. Proc.Natl. Acad. Sci. USA 106:16764–16769; 2009.

[2] Hannedouche, S.; Zhang, J.; Yi, T.; Shen, W.; Nguyen, D.; Pereira, J. P.; Guerini,D.; Baumgarten, B. U.; Roggo, S.; Wen, B.; Knochenmuss, R.; Noel, S.; Gessier,F.; Kelly, L. M.; Vanek, M.; Laurent, S.; Preuss, I.; Miault, C.; Christen, I.;Karuna, R.; Li, W.; Koo, D. I.; Suply, T.; Schmedt, C.; Peters, E. C.; Falchetto, R.;Katopodis, A.; Spanka, C.; Roy, M. O.; Detheux, M.; Chen, Y. A.; Schultz, P. G.;Cho, C. Y.; Seuwen, K.; Cyster, J. G. Sailer, A. W. Oxysterols direct immune cellmigration via EBI2. Nature 475:524–527; 2011.

[3] McDonald, J. G.; Russell, D. W. 25-Hydroxycholesterol: a new life inimmunology. J. Leukocyte Biol. 88:1071–1072; 2010.

W.J. Griffiths et al. / Free Radical Biology and Medicine 59 (2013) 69–84 83

[4] Goodwin, B.; Gauthier, K. C.; Umetani, M.; Watson, M. A.; Lochansky, M. I.;Collins, J. L.; Leitersdorf, E.; Mangelsdorf, D. J.; Kliewer, S. A.; Repa, J. J.Identification of bile acid precursors as endogenous ligands for the nuclearxenobiotic pregnane X receptor. Proc. Natl. Acad. Sci. USA 100:223–228; 2003.

[5] Janowski, B. A.; Grogan, M. J.; Jones, S. A.; Wisely, G. B.; Kliewer, S. A.; Corey,E. J.; Mangelsdorf, D. J. Structural requirements of ligands for the oxysterolliver X receptors LXRalpha and LXRbeta. Proc. Natl. Acad. Sci. USA 96:266–271;1999.

[6] Lehmann, J. M.; Kliewer, S. A.; Moore, L. B.; Smith-Oliver, T. A.; Oliver, B. B.;Su, J. L.; Sundseth, S. S.; Winegar, D. A.; Blanchard, D. E.; Spencer, T. A.;Willson, T. M. Activation of the nuclear receptor LXR by oxysterols defines anew hormone response pathway. J. Biol. Chem. 272:3137–3140; 1997.

[7] Nishimaki-Mogami, T.; Une, M.; Fujino, T.; Sato, Y.; Tamehiro, N.; Kawahara,Y.; Shudo, K.; Inoue, K. Identification of intermediates in the bile acidsynthetic pathway as ligands for the farnesoid X receptor. J. Lipid Res.45:1538–1545; 2004.

[8] Iuliano, L. Pathways of cholesterol oxidation via non-enzymatic mechanisms.Chem. Phys. Lipids 164:457–468; 2011.

[9] Brown, A. J.; Jessup, W. Oxysterols and atherosclerosis. Atherosclerosis142:1–28; 1999.

[10] Shentu, T. P.; Singh, D. K.; Oh, M. J.; Sun, S.; Sadaat, L.; Makino, A.; Mazzone,T.; Subbaiah, P. V.; Cho, M.; Levitan, I. The role of oxysterols in control ofendothelial stiffness. J. Lipid Res. 53:1348–1358; 2012.

[11] Leoni, V.; Caccia, C. Oxysterols as biomarkers in neurodegenerative diseases.Chem. Phys. Lipids 164:515–524; 2011.

[12] Burstein, S.; Middleditch, B. S.; Gut, M. Mass spectrometric study of theenzymatic conversion of cholesterol to (22R)-22-hydroxycholesterol,(20R,22R)-20,22-dihydroxycholesterol, and pregnenolone, and of (22R)-22-hydroxycholesterol to the glycol and pregnenolone in bovine adrenocorticalpreparations: mode of oxygen incorporation. J. Biol. Chem 250:9028–9037;1975.

[13] Russell, D. W. The enzymes, regulation, and genetics of bile acid synthesis.Annu. Rev. Biochem. 72:137–174; 2003.

[14] Dussault, I.; Yoo, H. D.; Lin, M.; Wang, E.; Fan, M.; Batta, A. K.; Salen, G.;Erickson, S. K.; Forman, B. M. Identification of an endogenous ligand thatactivates pregnane X receptor-mediated sterol clearance. Proc. Natl. Acad. Sci.USA 100:833–838; 2003.

[15] Nehring, J. A.; Zierold, C.; DeLuca, H. F. Lithocholic acid can carry out in vivofunctions of vitamin D. Proc. Natl. Acad. Sci. USA 104:10006–10009; 2007.

[16] Radhakrishnan, A.; Ikeda, Y.; Kwon, H. J.; Brown, M. S.; Goldstein, J. L. Sterol-regulated transport of SREBPs from endoplasmic reticulum to Golgi: oxyster-ols block transport by binding to Insig. Proc. Natl. Acad. Sci. USA104:6511–6518; 2007.

[17] Wang, Y.; Muneton, S.; Sjovall, J.; Jovanovic, J. N.; Griffiths, W. J. The effect of24S-hydroxycholesterol on cholesterol homeostasis in neurons: quantitativechanges to the cortical neuron proteome. J. Proteome Res. 7:1606–1614; 2008.

[18] Iguchi, Y.; Yamaguchi, M.; Sato, H.; Kihira, K.; Nishimaki-Mogami, T.; Une, M.Bile alcohols function as the ligands of membrane-type bile acid-activated Gprotein-coupled receptor. J. Lipid Res. 51:1432–1441; 2010.

[19] Liu, C.; Yang, X. V.; Wu, J.; Kuei, C.; Mani, N. S.; Zhang, L.; Yu, J.; Sutton, S. W.;Qin, N.; Banie, H.; Karlsson, L.; Sun, S.; Lovenberg, T. W. Oxysterols directB-cell migration through EBI2. Nature 475:519–523; 2011.

[20] Bensinger, S. J.; Bradley, M. N.; Joseph, S. B.; Zelcer, N.; Janssen, E. M.;Hausner, M. A.; Shih, R.; Parks, J. S.; Edwards, P. A.; Jamieson, B. D.; Tontonoz,P. LXR signaling couples sterol metabolism to proliferation in the acquiredimmune response. Cell 134:97–111; 2008.

[21] Diczfalusy, U.; Olofsson, K. E.; Carlsson, A. M.; Gong, M.; Golenbock, D. T.;Rooyackers, O. Flaring, U.; Bjorkbacka, H. Marked upregulation of cholesterol25-hydroxylase expression by lipopolysaccharide. J. Lipid Res. 50:2258–2264;2009.

[22] Dzeletovic, S.; Breuer, O.; Lund, E.; Diczfalusy, U. Determination of choles-terol oxidation products in human plasma by isotope dilution–mass spectro-metry. Anal. Biochem. 225:73–80; 1995.

[23] McDonald, J. G.; Thompson, B. M.; McCrum, E. C.; Russell, D. W. Extractionand analysis of sterols in biological matrices by high performance liquidchromatography electrospray ionization mass spectrometry. Methods Enzymol.432:145–170; 2007.

[24] McDonald, J. G.; Smith, D. D.; Stiles, A. R.; Russell, D. W. A comprehensivemethod for extraction and quantitative analysis of sterols and secosteroidsfrom human plasma. J. Lipid Res. 53:1399–1409; 2012.

[25] Quehenberger, O.; Armando, A. M.; Brown, A. H.; Milne, S. B.; Myers, D. S.;Merrill, A. H.; Bandyopadhyay, S.; Jones, K. N.; Kelly, S.; Shaner, R. L.; Sullards,C. M.; Wang, E.; Murphy, R. C.; Barkley, R. M.; Leiker, T. J.; Raetz, C. R.; Guan,Z.; Laird, G. M.; Six, D. A.; Russell, D. W.; McDonald, J. G.; Subramaniam, S.;Fahy, E.; Dennis, E. A. Lipidomics reveals a remarkable diversity of lipids inhuman plasma. J. Lipid Res. 51:3299–3305; 2010.

[26] DeBarber, A. E.; Sandlers, Y.; Pappu, A. S.; Merkens, L. S.; Duell, P. B.; Lear, S.R.; Erickson, S. K.; Steiner, R. D. Profiling sterols in cerebrotendinousxanthomatosis: utility of Girard derivatization and high resolution exactmass LC-ESI-MS(n) analysis. J. Chromatogr. B Anal. Technol. Biomed. Life Sci.879:1384–1392; 2011.

[27] Honda, A.; Yamashita, K.; Hara, T.; Ikegami, T.; Miyazaki, T.; Shirai, M.; Xu, G.;Numazawa, M.; Matsuzaki, Y. Highly sensitive quantification of key regula-tory oxysterols in biological samples by LC-ESI-MS/MS. J. Lipid Res. 50:350–357; 2009.

[28] Jiang, X.; Ory, D. S.; Han, X. Characterization of oxysterols by electrosprayionization tandem mass spectrometry after one-step derivatization withdimethylglycine. Rapid Commun. Mass Spectrom 21:141–152; 2007.

[29] Roberg-Larsen, H.; Strand, M.F.; Grimsmo, A.; Olsen, P.A.; Dembinski, J.L.;Rise, F.; Lundanes, E.; Greibrokk, T.; Krauss, S.; Wilson, S.R. High sensitivitymeasurements of active oxysterols with automated filtration/filterbackflush–solid phase extraction–liquid chromatography–mass spectrome-try. J. Chromatogr. A (in press); 2012.

[30] Meaney, S.; Heverin, M.; Panzenboeck, U.; Ekstrom, L.; Axelsson, M.;Andersson, U.; Diczfalusy, U.; Pikuleva, I.; Wahren, J.; Sattler, W.; Bjorkhem,I. Novel route for elimination of brain oxysterols across the blood–brainbarrier: conversion into 7alpha-hydroxy-3-oxo-4-cholestenoic acid. J. LipidRes. 48:944–951; 2007.

[31] Axelson, M.; Mork, B.; Sjovall, J. Occurrence of 3 beta-hydroxy-5-cholestenoicacid, 3 beta,7 alpha-dihydroxy-5-cholestenoic acid, and 7 alpha-hydroxy-3-oxo-4-cholestenoic acid as normal constituents in human blood. J. Lipid Res29:629–641; 1988.

[32] Brown, H. A.; Murphy, R. C. Working towards an exegesis for lipids in biology.Nat. Chem. Biol. 5:602–606; 2009.

[33] Dennis, E. A. Lipidomics joins the omics evolution. Proc. Natl. Acad. Sci. USA106:2089–2090; 2009.

[34] Gross, R. W.; Han, X. Shotgun lipidomics of neutral lipids as an enablingtechnology for elucidation of lipid-related diseases. Am. J. Physiol Endocrinol.Metab 297:E297–E303; 2009.

[35] Shevchenko, A.; Simons, K. Lipidomics: coming to grips with lipid diversity.Nat. Rev. Mol. Cell Biol. 11:593–598; 2010.

[36] Halket, J. M.; Zaikin, V. G. Derivatization in mass spectrometry. 1. Silylation.Eur. J. Mass Spectrom. (Chichester) 9:1–21; 2003.

[37] Fahy, E.; Subramaniam, S.; Brown, H. A.; Glass, C. K.; Merrill Jr A. H.; Murphy,R. C.; Raetz, C. R.; Russell, D. W.; Seyama, Y.; Shaw, W.; Shimizu, T.; Spener,F.; van, M. G.; VanNieuwenhze, M. S.; White, S. H.; Witztum, J. L.; Dennis, E. A.A comprehensive classification system for lipids. J. Lipid Res. 46:839–861;2005.

[38] Graessler, J.; Schwudke, D.; Schwarz, P. E.; Herzog, R.; Shevchenko, A.;Bornstein, S. R. Top-down lipidomics reveals ether lipid deficiency in bloodplasma of hypertensive patients. PLoS One 4(e6261); 2009.

[39] Honour, J.; Shackleton, C. H.; Griffiths, W. J.; Makin, H. L. General methods forthe extraction, purification, and measurement of steroids by chromatographyand mass spectrometry. In: Makin, H. L., Gower, D. B., editors. Steroid Analysis.New York: Springer; 2010. p. 163–282.

[40] Griffiths, W. J.; Liu, S.; Alvelius, G.; Sjovall, J. Derivatisation for the character-isation of neutral oxosteroids by electrospray and matrix-assisted laserdesorption/ionisation tandem mass spectrometry: the Girard P derivative.Rapid Commun. Mass Spectrom 17:924–935; 2003.

[41] Griffiths, W. J.; Wang, Y.; Alvelius, G.; Liu, S.; Bodin, K.; Sjovall, J. Analysis ofoxysterols by electrospray tandem mass spectrometry. J. Am. Soc. MassSpectrom 17:341–362; 2006.

[42] Griffiths, W. J.; Hornshaw, M.; Woffendin, G.; Baker, S. F.; Lockhart, A.;Heidelberger, S.; Gustafsson, M.; Sjovall, J.; Wang, Y. Discovering oxysterolsin plasma: a window on the metabolome. J. Proteome Res. 7:3602–3612;2008.

[43] Karu, K.; Hornshaw, M.; Woffendin, G.; Bodin, K.; Hamberg, M.; Alvelius, G.;Sjovall, J.; Turton, J.; Wang, Y.; Griffiths, W. J. Liquid chromatography–massspectrometry utilizing multi-stage fragmentation for the identification ofoxysterols. J. Lipid Res. 48:976–987; 2007.

[44] Karu, K.; Turton, J.; Wang, Y.; Griffiths, W. J. Nano-liquid chromatography–tandem mass spectrometry analysis of oxysterols in brain: monitoring ofcholesterol autoxidation. Chem. Phys. Lipids 164:411–424; 2011.

[45] Ogundare, M.; Theofilopoulos, S.; Lockhart, A.; Hall, L. J.; Arenas, E.; Sjovall, J.;Brenton, A. G.; Wang, Y.; Griffiths, W. J. Cerebrospinal fluid steroidomics: arebioactive bile acids present in brain? J. Biol. Chem. 285:4666–4679; 2010.

[46] Schroepfer Jr. G. J. Oxysterols: modulators of cholesterol metabolism andother processes. Physiol. Rev. 80:361–554; 2000.

[47] MacLachlan, J.; Wotherspoon, A. T.; Ansell, R. O.; Brooks, C. J. Cholesteroloxidase: sources, physical properties and analytical applications. J. SteroidBiochem. Mol. Biol. 72:169–195; 2000.

[48] Schwartz, J.C.; Syka, J.E.; Quarmby, S.T. 53rd ASMS Conference on MassSpectrometry and Allied Topics. 2005.

[49] Yuri, M.; Tokumoto, M.; Hara, N.; Fujimoto, Y.; Kobayashi, N.; Morisaki, M.Identification of 27-nor-3 alpha,7 alpha,12 alpha-trihydroxycoprostan-24-one apparently derived from 3 alpha,7 alpha,12 alpha-trihydroxy-24-oxoco-prostanoic acid, a postulated intermediate of bile acid biosynthesis. Chem.Pharm. Bull. (Tokyo) 41:1327–1329; 1993.

[50] Bun-ya, M.; Maebuchi, M.; Kamiryo, T.; Kurosawa, T.; Sato, M.; Tohma, M.;Jiang, L. L.; Hashimoto, T. Thiolase involved in bile acid formation. J. Biochem.123:347–352; 1998.

[51] Karlaganis, G.; Karlaganis, V.; Sjovall, J. Identification of 27-nor-5 beta-cholestane-3 alpha,7 alpha,12 alpha,24 xi, 25 xi,26-hexol and partial char-acterization of the bile alcohol profile in urine. J. Lipid Res 25:693–702; 1984.

[52] Griffiths, W. J.; Sjovall, J. Analytical strategies for characterization of bile acidand oxysterol metabolomes. Biochem. Biophys. Res. Commun. 396:80–84;2010.

[53] Axelson, M.; Sjovall, J. Potential bile acid precursors in plasma—possibleindicators of biosynthetic pathways to cholic and chenodeoxycholic acids inman. J. Steroid Biochem 36:631–640; 1990.

W.J. Griffiths et al. / Free Radical Biology and Medicine 59 (2013) 69–8484

[54] Lovgren-Sandblom, A.; Heverin, M.; Larsson, H.; Lundstrom, E.; Wahren, J.;Diczfalusy, U.; Bjorkhem, I.; Novel, L. C. -M. /MS method for assay of 7alpha-hydroxy-4-cholesten-3-one in human plasma: evidence for a significantextrahepatic metabolism. J. Chromatogr. B Anal. Technol. Biomed. Life Sci.856:15–19; 2007.

[55] Wong, J.; Quinn, C. M.; Gelissen, I. C.; Brown, A. J. Endogenous 24(S),25-epoxycholesterol fine-tunes acute control of cellular cholesterol homeostasis.J. Biol. Chem. 283:700–707; 2008.

[56] Wang, Y.; Sousa, K. M.; Bodin, K.; Theofilopoulos, S.; Sacchetti, P.; Hornshaw,M.; Woffendin, G.; Karu, K.; Sjovall, J.; Arenas, E.; Griffiths, W. J. Targetedlipidomic analysis of oxysterols in the embryonic central nervous system.Mol. Biosyst. 5:529–541; 2009.

[57] Mano, N.; Sato, Y.; Nagata, M.; Goto, T.; Goto, J. Bioconversion of 3beta-hydroxy-5-cholenoic acid into chenodeoxycholic acid by rat brain enzymesystems. J. Lipid Res. 45:1741–1748; 2004.

[58] Shoda, J.; Toll, A.; Axelson, M.; Pieper, F.; Wikvall, K.; Sjovall, J. Formation of7 alpha- and 7 beta-hydroxylated bile acid precursors from 27-hydroxycho-lesterol in human liver microsomes and mitochondria. Hepatology 17:395–403; 1993.

[59] Bjorkhem, I.; Hansson, M. Cerebrotendinous xanthomatosis: an inborn errorin bile acid synthesis with defined mutations but still a challenge. Biochem.Biophys. Res. Commun. 396:46–49; 2010.

[60] Clayton, P. T.; Verrips, A.; Sistermans, E.; Mann, A.; Mieli-Vergani, G.; Wevers,R. Mutations in the sterol 27-hydroxylase gene (CYP27A) cause hepatitis ofinfancy as well as cerebrotendinous xanthomatosis. J. Inherit. Metab. Dis.25:501–513; 2002.

[61] Clayton, P. T. Disorders of bile acid synthesis. J. Inherit. Metab. Dis. 34:593–604; 2011.

[62] Egestad, B.; Pettersson, P.; Skrede, S.; Sovall, J. Fast atom bombardment massspectrometry in the diagnosis of cerebrotendinous xanthomatosis. Scand.J. Clin. Lab. Invest. 45:443–446; 1985.

[63] Kuuranne, T.; Vahermo, M.; Leinonen, A.; Kostianen, R. Electrospray andatmospheric pressure chemical ionization tandem mass spectrometric beha-vior of eight anabolic steroid glucuronides. J. Am. Soc. Mass Spectrom11:722–730; 2000.

[64] Hoshita, T.; Yasuhara, M.; Une, M.; Kibe, A.; Itoga, E.; Kito, S.; Kuramoto, T.Occurrence of bile alcohol glucuronides in bile of patients with cerebroten-dinous xanthomatosis. J. Lipid Res. 21:1015–1021; 1980.